An Empirical Study of Early Warning Model on the Number of Coal Mine Accidents in China

Qing Liu,Jian Liu,Jinxin Gao,Jingjing Wang,Jing Han
DOI: https://doi.org/10.1016/j.ssci.2019.104559
IF: 6.392
2019-01-01
Safety Science
Abstract:Coal mining is an important energy industry as well as high risk industry with high accident rate. The objective of this study is to establish an early warning model for coal mine accidents. Based on the accident data in the recent 11 years, it is concluded that the coal mine accidents data has the characteristics of large fluctuations, small sample size, small value and large numerical nonlinear characteristics. After literature consulting, data analysis and variable comparison, Policy intervention degree (PID), Manufacturing Purchasing Managers' Index (PMI), Producer Price Index for Industrial Products (PPI), Main raw material purchase price index (RMPPI), Employment index (ED are selected as the auxiliary variables to construct the early warning model. VAR model is applied to determine the model structure. Results show that BP neural network model is suitable for the prediction of the number of coal mine accident.
What problem does this paper attempt to address?